This repository contains the code and results of the evaluation of our fuzzing tool. It was part of a project at the TU Delft, with four other members: Melchior Oudemans, Lars Rhijnsburger, Bo van den Berg, Martijn Smits. The code on this repository is an extension of the BigFuzz repository: https://github.com/qianzhanghk/BigFuzz. The results can be found in the dataset folder, the java files for running the benchmarks in the CustomArray folder. There is also a folder called benchmarks, which contains the original spark code of the benchmarks. Our extension can be found in the following folder: \fuzz\src\main\java\edu\tud\cs\jqf\bigfuzzplus\systematicMutation
To reproduce our research run the tool with the following configuration:
- SDK 8
- Module: bigfuzz-dfcoverage
- Main method class: BigFuzzPlusDriver The following program arguments can be given:
- Benchmark driver class.
- Benchmark test method in driver class.
- Mutation testing approach. Should be set to SystematicMutation, other options were made by colleagues and are not supported for this repository.
- Number of trials.
- Whether to exhaustively explore all columns.
- Mutation tree depth.
Example for SalaryAnalysis benchmark P11:
- SalaryAnalysisDriver testSalaryAnalysis SystematicMutation 10000 true 4
For an explanation of the parameters, see the accompanying research paper.
Author: Lars van Koetsveld van Ankeren
Supervisor: Burcu Kulahcioglu Ozkan
Technical University Delft.